Towards Consistency Checking Between Software Architecture and Informal Documentation

In the development process, documenting the software architecture is important to capture all reasoning and design decisions. Without a good and complete documentation, there is a lot of tacit knowledge that easily can get lost resulting in threats for success and increased costs. However, software architecture documentation is often missing or outdated. One reason for it is the tedious and costly process of creating and updating documentation in comparison to (perceived) low benefits. In this paper, we first present our long-term vision, where all information from any sources are persisted to avoid losing crucial information about a system. A base problem in this vision is keeping information from different sources consistent, with a major challenge of keeping consistency between models and informal documentation. We plan to address checking the consistency between models and textual natural language artefacts using natural language understanding. For this, we plan to break down the task into smaller subtasks that should provide a better understanding of the documents and their semantics. The extracted information should then be used to create traceability links and to check whether statements within the textual documentation are consistent with the software architecture models.